Corporate Deep Learning with Caffe Training Course

1. What is deep learning?

1. Definition and concept of deep learning
2. Comparison with traditional machine learning algorithms
3. Key components and architecture of deep neural networks

2. Basics of a neural network

1. Structure of a neural network: Input layer, hidden layers, output layer
2. Neuron activation function: Sigmoid, and ReLU
3. Forward propagation: How inputs are processed through the network to generate outputs?

3. Working on a neural network

1. Backpropagation algorithm: Adjusting weights and biases to minimize error
2. Optimization algorithms: Gradient descent, stochastic gradient descent, mini-batch gradient descent
3. Training process: Splitting data into training, validation, and test sets

4. Gradient descent

1. Concept of gradient descent for optimizing neural network parameters
2. Understanding the cost function and its role in gradient descent
3. Types of gradient descent: Batch, Mini-batch, and Stochastic gradient descent

5. Image classification - Overview and application

1. Introduction to image classification tasks
2. Importance of image classification in various fields such as healthcare, autonomous vehicles, and security
3. Overview of image classification techniques: traditional methods vs. deep learning approaches

1. Basics of computer vision

1. Introduction to computer vision and its applications
2. Image representation: Pixels, color channels, and image formats
3. Image preprocessing techniques: Normalization, resizing, and augmentation

2. What are convolutional neural networks?

1. Evolution of CNNs from traditional neural networks
2. Role of convolutional layers in feature extraction
3. Understanding filter kernels and feature maps

3. CNN: Convolutions

1. Performing convolutions on input images
2. Stride and padding: controlling output size and boundary effects
3. Visualizing convolutional filters and feature maps

4. CNN: Activation and pooling

1. Activation functions in CNN layers: ReLU, Leaky ReLU, and others
2. Pooling layers: Max pooling, average pooling, and their benefits
3. Reducing spatial dimensions while preserving important features

5. CNN: Classification

1. Final layers in a CNN: Fully connected layers and output layer
2. Softmax function for multi-class classification
3. Training a CNN for image classification tasks